INDEX

Explanations

No Explanations Found

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ATK

-0.70

mort

-0.69

ammu

-0.67

 Aging

-0.67

ELY

-0.66

AM

-0.66

 Reincarn

-0.65

 PEOPLE

-0.65

aval

-0.64

phrine

-0.64

POSITIVE LOGITS

ful

0.70

chery

0.65

imeters

0.65

toe

0.65

 todd

0.63

 reel

0.62

hole

0.62

âĸł

0.61

 improv

0.61

 notch

0.61

Activations Density 0.000%

No Known Activations

This feature has no known activations.